AI, Court Cases, and “Human Authorship -

As our government struggles to deal with the implications of artificial intelligence in the wake of popular awareness and daily application, there are several important issues being debated within AI: what can it be trained on, what can it produce, and is there enough “human authorship” in generative AI that is removed from the automated process for any human to claim a copyright in any end product?

To put a finer point on the third question, we first have to look at the concept of “human authorship” within the context of the Copyright Act of 1976, as well as its forerunners.

The Copyright Act does not explicitly require that the author be human within the text of the law. However, the Supreme Court elaborated that copyright law protects “the fruits of intellectual labor” that “are founded in the creative powers of the mind,” *Trade-Mark Cases,* 100 U.S. 82, 94(1879).

Burrow-Giles

A mere 5 years after this declaration, *Burrow-Giles* made its way up to the Supremes. *Burrow-Giles Lithograph Co. v. Sarony *addressed the issue of the threshold of human authorship in light of new technology when a lithographer (Burrow-Giles) was making copies of a photo of Oscar Wilde without the copyright held by the Sarony side. At issue was whether a photographer has a copyright to their photo and can thus claim infringement

when their photos are copied, i.e. by a lithographer – seemingly an open-and-shut case, but with an important concept to consider in additional nuance: what satisfies the requirement of human authorship?

While we may find it obvious that a photographer would have the rights to their photo today, in the 1800s, photography was still a new art form with a body of caselaw that was just starting to develop around it. The largest source of nuance to the arguments is the automated process that a camera performs between the photographer opening the shutter and the photo being produced, which is a complex mechanical and chemical reaction that is free from human interference once begun. The Burrow-Giles side argued that, because this process is automated, it separates the final product of the negatives (or whatever this process made at the time) used to produce a photo from the authorship of the human behind the camera who would claim the copyright. This is unexpectedly similar to the case of *Naruto v. Slater*, in which a macaque monkey fiddled around with a camera in a moment of monkey-see-monkey-do and took a selfie. Naturally, PETA sued on the monkey’s behalf, claiming it had the copyright to the photo that was infringed when it was not compensated.

Skipping to the end, in *Burrow-Giles*, the court held that, while there is an automated process that would not in and of itself qualify for copyright protection as it is removed from human authorship, posing Oscar Wilde, choosing his outfit, fixing the shutter speed, and controlling other factors of the photo was enough human control of the final product to grant said human a copyright.

In *Naruto*, PETA claimed Naruto, the clever little photo-monkey, suffered economic harm as a result of copyright infringement when Slater and Wildlife Personalities ran the “monkey selfie” photo. The court in the 9th Circuit found that, while the monkey DID have constitutional standing, it did not have statutory standing as the Copyright Act of 1976 protects works of human authorship, not that of animals.

How does this create a throughline to today?

Between the entering of a prompt and the final product in most generative AI is a process that is just as automated as clicking the shutter and prompting the camera’s mechanisms and chemical reaction in *Burrow-Giles*. Similarly, AI, much like the macaque, is not a human and, at least until we go full *Black Mirror* when AI becomes uncomfortably self-aware and may get more rights, the program is not entitled to copyright protection.

Keep in mind that there is still a *LOT* of ambiguity and room for argument: when a very elaborate prompt is created, while the prompt will likely qualify for copyright protection in and of itself, would it eventually, be enough human authorship (similar to posing Oscar Wilde and choosing the lighting) to create a copyright in the end product? Is the training of artificial intelligence on copyrighted materials enough to claim infringement, an issue currently making its way up the courts in*Anderson v. Stability AI Ltd* within the 9th Circuit? Will Skynet treat us nicely and give us all AI spouses?

While there are challenges, i.e. *Thaler v. Perlmutter *in which Thaler is claiming that the Copyright Act does not require “human” authorship as elaborated by the copyright office and *Burrow-Giles*, copyright nerds such as myself are waiting with bated breath to see how the “YOLO court” of today may choose to address the issue in light of Congressional inaction.